Methods of making single-stranded circular oligonucleotides

Description Submit all comments and votes

FIELD OF THE INVENTION

The present invention provides single-stranded circular oligonucleotides capable of binding to a target DNA or RNA and thereby regulating DNA replication, RNA transcription, protein translation, and other processes involving nucleic acid templates. Furthermore, circular oligonucleotides can be labeled for use as probes to detect or isolate a target nucleic acid. Circular oligonucleotides can also displace one strand of a duplex nucleic acid without prior denaturation of the duplex. Moreover, circular-oligonucleotides are resistant to exonucleases and bind to a target with higher selectivity and affinity than do linear oligonucleotides.

BACKGROUND OF THE INVENTION

An oligonucleotide binds to a target nucleic acid by forming hydrogen bonds between bases in the target and the oligonucleotide. Common B DNA has conventional adenine-thymine (A-T), and guanine-cytosine (G-C) Watson and Crick base pairs with two and three hydrogen bonds, respectively. Conventional hybridization technology is based upon the capability of sequence-specific DNA or RNA probes to bind to a target nucleic acid via Watson-Crick hydrogen bonds. However, other types of hydrogen bonding patterns are known wherein some atoms of a base which are not involved in Watson-Crick base pairing can form hydrogen bonds to another nucleotide. For example, thymine (T) can bind to an A-T Watson-Crick base pair via hydrogen bonds to the adenine, thereby forming a T-AT base triad. Hoogsteen (1959, Acta Crystallography 12: 822) first described the alternate hydrogen bonds present in T-AT and C-GC base triads. More recently, G-TA base triads, wherein guanine can hydrogen bond with a central thymine, have been observed(Griffin et al., 1989, Science 245: 967-971). If an oligonucleotide could bind to a target with both Watson-Crick and alternate hydrogen bonds an extremely stable complex would form that would have a variety of in vivo and in vitro utilities. However, to date there has been no disclosure of an oligonucleotide with the necessary structural features to achieve stable target binding with both Watson-Crick and alternate hydrogen bonds.

Oligonucleotides have been observed to bind by non-Watson-Crick hydrogen bonding in vitro. For example, Cooney et al., 1988, Science 241:456 disclose a 27-base single-stranded oligonucleotide which bound to a double-stranded nucleic acid via non-Watson-Crick hydrogen bonds. However, triple-stranded complexes of this type are not very stable, because the oligonucleotide is bound to its target only with less stable alternate hydrogen bonds, i.e., without any Watson-Crick bonds.

Oligonucleotides have been used for a variety of utilities. For example, oligonucleotides can be used as probes for target nucleic acids that are immobilized onto a filter or membrane, or are present in tissues. Sambrook et al. (1989, Molecular Cloning: A Laboratory Manual, Vols. 1-3, Cold Spring Harbor Press, N.Y.) provide a detailed review of hybridization techniques.

Furthermore, there has been great interest recently in developing oligonucleotides as regulators of cellular nucleic acid biological function. This interest arises from observations on naturally occurring complementary, or antisense, RNA used by some cells to control protein expression. However, the development of oligonucleotides for in vivo regulation of biological processes has been hampered by several long-standing problems, including the low binding stability and nuclease sensitivity of linear oligonucleotides.

For example, transcription of the human c-myc gene has been inhibited in a cell free, in vitro assay system by a 27-base linear oligonucleotide designed to bind to the c-myc promoter. Inhibition was only observed using a carefully controlled in vitro assay system wherein lower than physiological temperatures were employed, and many cellular enzymes had been removed or inactivated. These conditions were necessary because linear oligonucleotides bind with low affinity and are highly susceptible to enzymes which degrade linear pieces of DNA (Cooney et al.). Splicing of a pre-mRNA transcript essential for Herpes Simplex virus replication has also been inhibited with a linear oligonucleotide which was complementary to an acceptor splice junction. In this instance, a methylphosphonate linkage was employed in the linear oligonucleotide to increase its nuclease resistance. Addition of this chemically-modified oligonucleotide to the growth medium caused reduction in protein synthesis and growth of uninfected cells, most likely because of toxicity problems at high concentrations (Smith et al., 1986, Proc. Natl. Acad. Sci. USA 83: 2787-2791).

In another example, linear oligonucleotides were used to inhibit human immunodeficiency virus replication in cultured cells. Linear oligonucleotides complementary to sites within or near the terminal repeats of the retrovirus genome and within sites complementary to certain splice junctions were most effective in blocking viral replication. However, these experiments required large amounts of the linear oligonucleotides before an effect was obtained, presumably because of the low binding stability and vulnerability of these linear oligonucleotides to nucleases (Goodchild et al., 1988, Proc. Natl. Acad. Sci. USA 85: 5507-5511).

Accordingly, oligonucleotides that are useful as regulators of biological processes preferably possess certain properties. First, the oligonucleotide should bind strongly enough to its complementary target nucleic acid to have the desired regulatory effect. Second, it is generally desirable that the oligonucleotide and its target be sequence specific. Third, the oligonucleotide should have a sufficient half-life under in vivo conditions for it to be able to accomplish its desired regulatory action in the cell. Hence, the oligonucleotide should be resistant to enzymes that degrade nucleic acids, e.g. nucleases. Fourth, the oligonucleotide should be able to bind to single- and double-stranded targets.

While linear oligonucleotides may satisfy the requirement for sequence specificity, linear oligonucleotides are sensitive to nucleases and generally require chemical modification to increase biological half-life. Such modifications increase the cost of making an oligonucleotide and may present toxicity problems. Furthermore, linear oligonucleotides bind to form a two-stranded complex like those present in cellular nucleic acids. Consequently, cellular enzymes can readily manipulate and dissociate a linear oligonucleotide bound in a double-stranded complex with target. The low binding strength and nuclease sensitivity of linear oligonucleotides can thus necessitate administration of high concentrations of oligonucleotide, in turn making such administration toxic or costly. Moreover, while linear oligonucleotides can bind to a double-stranded target via alternate hydrogen bonds (e.g. Hoogsteen binding), linear oligonucleotides cannot readily dissociate a double-stranded target to replace one strand and thereby form a more stable Watson-Crick bonding pattern.

Furthermore, increased binding strength increases the effectiveness of a regulatory oligonucleotide. Therefore, an oligonucleotide with high binding affinity can be used at lower dosages. Lower dosages decrease costs and reduce the likelihood that a chemically-modified oligonucleotide will be toxic. Therefore, high oligonucleotide binding affinity for target is a highly desirable trait.

Accordingly, the present invention provides single-stranded circular oligonucleotides which, by nature of the circularity of the oligonucleotide and the domains present on the oligonucleotide, are nuclease resistant and bind with strong affinity and high selectivity to their targeted nucleic acids. Moreover, the present circular oligonucleotides can dissociate and bind to a double-stranded target without prior denaturation of that target.

Some types of single-stranded circles of DNA or RNA are known. For example, the structures of some naturally occurring viral and bacteriophage genomes are single-stranded circular nucleic acids. Single-stranded circles of DNA have been studied by Erie et al. (1987, Biochemistry 26:7150-7159 and 1989, Biochemistry 28: 268-273). However, none of these circular molecules are designed to bind a target nucleic acid. Hence, the present invention represents an innovation characterized by a substantial improvement relative to the prior art since the subject circular oligonucleotides exhibit high specificity, low or no toxicity and more resistance to nucleases than linear oligonucleotides, while binding to single- or double-stranded target nucleic acids more strongly than conventional linear oligonucleotides.

SUMMARY OF THE INVENTION

The present invention provides a single-stranded circular oligonucleotide having at least one parallel binding (P) domain and at least one anti-parallel binding (AP) domain, and having a loop domain between each binding domain to form the circular oligonucleotide. Each P and corresponding AP domain has sufficient complementarity to bind detectably to one strand of a defined nucleic acid target with the P domain binding in a parallel manner to the target, and the AP domain binding in an anti-parallel manner to the target. Sufficient complementarity means that a sufficient number of base pairs exists between the target nucleic acid and the P and/or AP domains of the circular oligonucleotide to achieve stable, i.e. detectable, binding.

In the case where multiple P and AP binding domains are included in the circular oligonucleotides of the present invention, the loop domains separating the P and AP binding domains can constitute, in whole or in part, another P or AP domain which functions as a binding domain in an alternate conformation. In other words, depending upon the particular target, a binding domain (P or AP) can also function as a loop domain for another binding domain and vice versa.

Another aspect of the present invention provides the subject single-stranded circular oligonucleotides derivatized with a reporter molecule to provide a probe for a target nucleic acid, or with a drug or other pharmaceutical agent to provide cell specific drug delivery, or with agents which can cleave or otherwise modify the target nucleic acid or, furthermore, with agents that can facilitate cellular uptake or target binding of the oligonucleotide.

An additional aspect of the present invention provides single-stranded circular oligonucleotides linked to a solid support for isolation of a nucleic acid complementary to the oligonucleotide.

Another aspect of the present invention provides a compartmentalized kit for detection or diagnosis of a target nucleic acid including at least one first container providing any one of the present circular oligonucleotides.

A further aspect of the present invention provides a method of detecting a target nucleic acid which involves contacting a single-stranded circular oligonucleotide with a sample containing the target nucleic acid, for a time and under conditions sufficient to form an oligonucleotide-target complex, and detecting the complex. This detection method can be by fluorescent energy transfer.

A still further aspect of the present invention provides a method of regulating biosynthesis of a DNA, an RNA or a protein. This method includes contacting at least one of the subject circular oligonucleotides with a nucleic acid template for the DNA, the RNA or the protein under conditions sufficient to permit binding of the oligonucleotide to a target sequence contained in the template, followed by binding of the oligonucleotide to the target, blocking access to the template and thereby regulating biosynthesis of the DNA, the RNA or the protein.

An additional aspect of the present invention provides pharmaceutical compositions for regulating biosynthesis of a nucleic acid or protein containing a biosynthesis regulating amount of at least one of the subject circular oligonucleotides and a pharmaceutically acceptable carrier.

A further aspect of the present invention provides a method of preparing a single-stranded circular oligonucleotide which includes binding a linear precircle to an end-joining-oligonucleotide, joining the two ends of the precircle and recovering the circular oligonucleotide product.

Another aspect of the present invention provides a method of strand displacement in a double-stranded nucleic acid target by contacting the target with any one of the present circular oligonucleotides for a time and under conditions effective to denature the target and to bind the circular oligonucleotide.

DESCRIPTIONS OF THE DRAWINGS:

FIG. 1A depicts the bonding patterns of Watson-Crick (anti-parallel domain) AT and GC base pairs. FIG. 1B depicts T-AT, C+GC and G-TA base triads that can form between P, target and AP nucleotides.

FIG. 2 schematically illustrates a circularization reaction for synthesis of single-stranded circular oligonucleotides. A linear precircle oligonucleotide is bound to an oligonucleotide having the same sequence as the target, i.e. an end-joining-oligonucleotide, to form a precircle complex. After ligation, the circularized oligonucleotides are separated from the end-joining-oligonucleotide.

FIG. 3 depicts the sequence of linear precursors to circular oligonucleotides, i.e. precircles (1-3 having SEQ ID NO: 5, SEQ ID NO: 6 and SEQ ID NO: 7), targets (4,5 having SEQ ID NO: 8 and SEQ ID NO: 9), circular oligonucleotides (6,7,8 and 13 having SEQ ID NO: 5-7 and 14), and linear oligonucleotides (9-12 and 14 having SEQ ID NO: 10-13 and 15) described in the examples.

FIG. 4 depicts the structure of a linear precircle complexed with an end-joining-oligonucleotide before ligation.

FIG. 5 depicts the effect of pH on circular oligonucleotide:target complex formation as measured by Tm. Filled circles represent the stability at different pH values for a 6:4 complex while filled squares depict the stability of a 7:5 complex. The sequences of circular oligonucleotides 6 and 7 and targets 4 and 5 are presented in FIG. 3.

FIG. 6A depicts the effect of loop size on complex formation, with a comparison between binding to two targets: a simple (dA).sub.12 target (squares) and a 36 nucleotide oligonucleotide target (circles). FIG. 6B depicts the effect of target and binding domain length on complex formation.

FIG. 7 depicts a complex formed between a circular oligonucleotide and a target where the P and AP binding domains are staggered on the target.

FIG. 8 depicts replacement of one strand of a fluorescently labeled double stranded target (SEQ ID NO: 11) by either a linear oligonucleotide having SEQ ID NO: 8 (dotted line) or a circular oligonucleotide having SEQ ID NO: 5 (solid line). Strand replacement was measured by an increase in fluorescein fluorescence intensity (Y-axis) as a function of time (X-axis).

FIG. 9 depicts a plot of observed pseudo-first rate constant, K.sub.obs for duplex target (SEQ ID NO: order rate constant, K.sub.obs for duplex target (SEQ ID NO: 5) at several concentrations. Uncertainty in rate constants are no more than .+-.10%. The depicted curve is a rectangular hyperbola generated as a best fit. A double reciprocal plot of the data, i.e., [circular oligonucleotide].sup.-1 vs (K.sub.obs).sup.-1 is linear with a slope of 8.95.times.10.sup.-6 sec.M.sup.-1 and a y-intercept of 39.8 sec.

FIG. 10A depicts plots of the observed hyperchromicity (at 260 nm) as the temperature is increased for a circular oligonucleotide having two sets of binding domains and SEQ ID NO:18 when bound to either a target oligonucleotide having SEQ ID NO:19 (open circles) or to a target oligonucleotide having SEQ ID NO:20 (filled circles). These data indicate the melting temperature (T.sub.m) of the SEQ ID NO:18-SEQ ID NO:19 complex is 44.5.degree. C. and the T.sub.m of the SEQ ID NO:18-SEQ ID NO:20 complex is 47.5.degree. C.

FIG. 10B depicts the mole fraction present of the (SEQ ID NO:18) circular oligonucleotide having two pairs of binding domains versus the absorbance, when mixed with the SEQ ID NO:19 target (squares), the SEQ ID NO:20 target (triangles) or when mixed with a 1:1 combination both SEQ ID NO:19 and SEQ ID NO:20 targets (circles). The inflection point of the observed absorbance provides the mole fraction of SEQ ID NO:18 circular oligonucleotide needed for complete complexation with the indicated target oligonucleotides.

FIG. 11A is a schematic diagram illustrating the binding of a SEQ ID NO:18 circular oligonucleotide having two pairs of binding domains, i.e. I and II, with either of target oligonucleotide SEQ ID NO:19 or target oligonucleotide SEQ ID NO:20. This figure illustrates that when binding domain pair I has bound its target oligonucleotide, the P and AP domains of pair II serve as loop domains separating the P and AP binding domains of pair I, and vice versa.

FIG. 11B is a schematic diagram illustrating the effect of pH upon target selection by the SEQ ID NO:18 circular oligonucleotide which has two pairs of binding domains, i.e. I and II. In this case two target sites, complementary to the pair I and pair II binding domains, are present within a single oligonucleotide. When the pH is low, pair I binding domains which contain cytosine, preferentially bind to their complementary target, while the pair II binding domains which contain no cytosine, do not bind their target. However, when the pH is high, pair II binding domains containing no cytosine, preferentially bind to their target while the pair I binding domains remain unbound.

FIG. 12 depicts the melting temperature (T.sub.m) as a function of pH when the two binding domain SEQ ID NO:18 circular oligonucleotide is bound to target oligonucleotide SEQ ID NO:20 (open circles), SEQ ID NO:19 (open squares) or SEQ ID NO:21 (filled circles). Oligonucleotides having SEQ ID NO:19 or SEQ ID NO:20 had a single target for the SEQ ID NO:18 circular oligonucleotide, however the oligonucleotide having SEQ ID NO:21 encoded two separate target sites for the SEQ ID NO:18 circular oligonucleotide.

FIG. 13A depicts the absorbance versus mole fraction of SEQ ID NO:18 circular oligonucleotide present in a mixture with the longer two-target site oligonucleotide having SEQ ID NO:21. The mole fraction of circular oligonucleotide at complete complexation (inflection point in the observed absorbance) is about 0.63.

FIG. 13B depicts the observed T.sub.m values for the SEQ ID NO:18 circular oligonucleotide bound to the two target-site oligonucleotide having SEQ ID NO:21. As shown, there were two T.sub.m values at each of the pH values tested. These two T.sub.m values correspond to separate melting events at each of the two target sites within the SEQ ID NO:21 oligonucleotide.

FIG. 14A depicts the relative absorbance at 260 nm of increasing amounts of the SEQ ID NO:18 circular oligonucleotide bound to the two-target site SEQ ID NO:21 oligonucleotide at pH 5.5. The SEQ ID NO:21 oligonucleotide was present at 1.5 .mu.M and the SEQ ID NO:18 circular oligonucleotide concentration was present at 0, 0.25, 0.5, 1.0 and 2.0 molar equivalents (lower to upper curves, respectively). The temperature at which the absorbance increases dramatically corresponds to the melting temperature. Only one sharp increase in absorbance was observed at about 60.degree. C. when the circular oligonucleotide was present at 0, 0.25, 0.5 and 1.0 molar equivalents (lower four curves). However, two sharp increases in absorbance were observed at about 47.degree. C. and about 60.degree. C. when 2.0 molar equivalents of circular oligonucleotide were mixed with 1.0 molar equivalents of the SEQ ID NO:21 oligonucleotide.

FIG. 14B depicts the relative absorbance at 260 nm of increasing amounts of the SEQ ID NO:18 circular oligonucleotide bound to the two-target site SEQ ID NO:21 oligonucleotide at pH 8.5. The SEQ ID NO:21 oligonucleotide was present at 1.5 .mu.M and the SEQ ID NO:18 circular oligonucleotide concentration was present at 0, 0.25, 0.5, 1.0 and 2.0 molar equivalents (lower to upper curves, respectively). The observed melting points at low molar ratios of circular oligonucleotide to SEQ ID NO:21 oligonucleotide is about 52.degree. C. (FIG. 14B middle three curves middle three curves, corresponding to molar ratios of SEQ ID NO:18 to SEQ ID NO:21 oligonucleotide of 0.25, 0.5 and 1.0).

FIG. 15 depicts the hyperchromicity at pH 5.5 of a mixture of circular oligonucleotide (SEQ ID NO:18 at 1.5 .mu.M) with two-target site oligonucleotide (SEQ ID NO:21 at 1.5 .mu.M) in the presence of oligonucleotides having either SEQ ID NO:22 (TCTCTCTCT at 1.5 .mu.M, filled circles) or SEQ ID NO:23 (TTTTTTTTT at 1.5 .mu.M, open circles). Two inflections in hyperchromicity (open circles) indicate that binding has occurred at both target sites within the SEQ ID NO:21 oligonucleotide, whereas a single inflection (filled circles) indicates binding has occurred at only one site in the SEQ ID NO:21 oligonucleotide.

FIG. 16 provides a schematic diagram illustrating the circular arrangement of one set of P and AP domains relative to each other as well as when bound to a target strand (T). The arrows indicate the 5' to 3' orientation of each strand where the 5' end of each domain is the tail and the 3' end is the arrowhead.

DETAILED DESCRIPTION OF THE INVENTION

The present invention relates to single-stranded circular oligonucleotides, i.e. circles, which can bind to nucleic acid targets with higher affinity and selectivity than a corresponding linear oligonucleotide. Moreover, since the present circles can open up two strands of a double-stranded nucleic acid and bind thereto, both single- and double-stranded nucleic acids can be targets for binding by the present circular oligonucleotides.

Furthermore, the strong, selective binding of these circles to either single- or double-stranded targets provides a variety of uses, including methods of regulating such biological processes as DNA replication, RNA transcription, RNA splicing and processing, protein translation and the like. Similarly, the ability of these circles to dissociate double-stranded nucleic acids and to selectively and stably bind to targeted nucleic acids makes them ideal as diagnostic probes or as markers to localize, for example, specific sites in a chromosome or other DNA or RNA molecules. Additionally, the present circles are useful for isolation of complementary nucleic acids or for sequence-specific delivery of drugs or other molecules into cells.

In particular, the single-stranded circular oligonucleotides of the present invention have at least one parallel binding (P) domain and at least one anti-parallel binding (AP) domain and have a loop domain between each binding domain, so that a circular oligonucleotide is formed. Moreover, each P and AP domain exhibits sufficient complementarity to bind to one strand of a defined nucleic acid target with the P domain binding to the target in a parallel manner and the AP domain binding to the target in an anti-parallel manner.

The schematic illustration set forth in FIG. 16 shows the circular arrangement of one set of P and AP oligonucleotide domains relative to each other as well as when bound to a target (T, as indicated in FIG. 16).

The arrows indicate the 5' to 3' orientation of each strand with the 5' end of each domain at the tail and the 3' end at the arrowhead. Hence as used herein binding of nucleic acids in a parallel manner means that the 5' to 3' orientation is the same for each strand or nucleotide in the complex. This is the type of binding present between the target and the P domain. As used herein, binding of nucleic acids in an anti-parallel manner means that the 5' to 3' orientations of two strands or nucleotides in a complex lie in opposite directions, i.e. the strands are aligned as found in the typical Watson-Crick base pairing arrangement of double helical DNA.

When more than one P and AP binding domain is present, such binding domains are separated from other P and AP domains by loop domains whose lengths are sufficient to permit binding to multiple targets. Moreover, when a circular oligonucleotide has multiple AP and P domains, a loop domain for one pair of corresponding AP and P binding domains can constitute an AP or P domain for binding to another target. When a circular oligonucleotide of the present invention includes, e.g., two pairs of corresponding binding domains, these pairs of corresponding binding domains can also bind separate target sites. Moreover, when a circle has multiple AP and P domains, the corresponding targets need not be linked on one nucleic acid strand. Furthermore, a loop domain of a circular oligonucleotide bound to a given target can be an AP or P domain for binding to a second target when the circular oligonucleotide releases from the first target.

In accordance with this invention, the nucleotide sequences of the P and AP domains can be determined from the defined sequence of the nucleic acid target by reference to the base pairing rules provided hereinbelow. A target can be either single- or double-stranded and is selected by its known functional and structural characteristics. For example, some preferred targets can be coding regions, origins of replication, reverse transcriptase binding sites, transcription regulatory elements, RNA splicing junctions, or ribosome binding sites, among others. A target can also be selected by its capability for detection or isolation of a DNA or RNA template. Preferred targets are rich in purines, i.e. in adenines and guanines.

The nucleotide sequence of the target DNA or RNA can be known in full or in part. When the target nucleotide sequence is completely known the sequences of the P and AP domains are designed with the necessary degree of complementarity to achieve binding, as detected by known procedures, for example by a change in light absorption or fluorescence. In some instances, the target sequence can be represented by a consensus sequence or be only partially known. For example, circular oligonucleotides (circles) which bind to an entire class of targets represented by a consensus sequence can be provided by designing the P and AP domains from the target consensus sequence. In this instance some of the targets may match the consensus sequence exactly and others may have a few mismatched bases, but not enough mismatch to prevent binding. Likewise, if a portion of a target sequence is known, one skilled in the art can refer to the base pairing rules provided hereinbelow to design circles which bind to that target with higher affinity than a linear oligonucleotide that has a sequence corresponding to that of the circle.

Thus, the present invention is also directed to circles having P and AP domains which are sufficiently complementary to bind to a nucleic acid target wherein a sufficient number, but not necessarily all, nucleotide positions in the P and AP domains are determined from the target sequence in accordance with the base pairing rules of this invention. The number of determined (i.e. known) positions is that number of positions which are necessary to provide sufficient complementarity for binding of the subject oligonucleotides to their targets, as detected by standard procedures including a change in light absorption upon binding or melting.

The base pairing rules of the present invention provide for the P domain to bind to the target by forming base pairs wherein the P domain and target nucleotides have the same 5' to 3' orientation. In particular, these rules are satisfied to the extent needed to achieve binding of a circular oligonucleotide to its nucleic acid target, i.e. the degree of complementarity need not be 100% so long as binding can be detected. Hence, the general rules for determining the sequence of the P domain are thus:

when a base for a position in the target is guanine or a guanine analog, then P has cytosine, or a suitable analog thereof, in a corresponding position;

when a base for a position in the target is adenine, or an adenine analog then P has thymine or uracil, or suitable analogs thereof, in a corresponding position;

when a base for a position in the target is thymine, or a thymine analog, then P has cytosine or guanine, or suitable analogs thereof, in a corresponding position;

when a base for a position in the target is cytosine, or a cytosine analog, then P has cytosine, thymine or uracil, or suitable analogs thereof, in a corresponding position; and

when a base for a position in the target is uracil, or a uracil analog, then P has cytosine, guanine, thymine, or uracil, or suitable analogs thereof, in a corresponding position.

The base pairing rules of the present invention provide for the AP domain to bind to the target by forming base pairs wherein the AP domain and target nucleotides are oriented in opposite directions. In particular these rules are satisfied to the extent necessary to achieve detectable binding of a circular oligonucleotide to its nucleic acid target, i.e. the degree of complementarity can be less than 100%. Hence, the base pairing rules can be adhered to only insofar as is necessary to achieve sufficient complementarity for binding to be detected between the circular oligonucleotide and its target.

Thus, the general rules for determining the sequence of the AP domain are as follows:

when a base for a position in the target is guanine, or a guanine analog, then AP has cytosine or uracil, or suitable analogs thereof, in a corresponding position;

when a base for a position in the target is adenine, or an adenine analog, then AP has thymine or uracil, or suitable analogs thereof, in a corresponding position;

when a base for a position in the target is thymine, or a thymine analog, then AP has adenine, or a suitable analog thereof, in a corresponding position; and

when a base for a position in the target is cytosine, or a cytosine analog, then AP has a guanine, or a suitable analog thereof, in corresponding position;

when a base for a position in the target is uracil, or a uracil analog, then AP has adenine or guanine, or suitable analogs thereof, in a corresponding position.

In a preferred embodiment, the P, AP and loop domains are not complementary to each other.

Table 1 summarizes which nucleotides can form anti-parallel base pairs or parallel base pairs with a defined target nucleotide.

TABLE 1 ______________________________________ Target Anti-Parallel Parallel Domain Nucleotide.sup.a Domain Nucleotide.sup.a Nucleotide.sup.a ______________________________________ G C or U C A T or U T or U T A C or G C G C, T or U U A or G C, G, T or U ______________________________________ .sup.a or a suitable analog

Two complementary single-stranded nucleic acids form a stable double helix (duplex) when the strands bind, or hybridize, to each other in the typical Watson-Crick fashion, i.e. via anti-parallel GC and AT base pairs. For the present invention, stable duplex formation and stable triplex formation is achieved when the P and AP domains exhibit sufficient complementarity to the target sequence to achieve stable binding between the circular oligonucleotide and the target molecule. Stable binding occurs when an oligonucleotide remains detectably bound to target under the required conditions.

Complementarity between nucleic acids is the degree to which the bases in one nucleic acid strand can hydrogen bond, or base pair, with the bases in a second nucleic acid strand. Hence, complementarity can sometimes be conveniently described by the percentage, i.e. proportion, of nucleotides which form base pairs between two strands or within a specific region or domain of two strands. For the present invention sufficient complementarity means that a sufficient number of base pairs exist between a target nucleic acid and the P and/or AP domains of the circular oligonucleotide to achieve detectable binding. Moreover, the degree of complementarity between the P domain and the target and the AP domain and the target need not be the same. When expressed or measured by percentage of base pairs formed, the degree of complementarity can range from as little as about 30-40% complementarity to full, i.e. 100%, complementarity. In general, the overall degree of complementarity between the P or AP domain and the target is preferably at least about 50%. However, the P domain can sometimes have less complementarity with the target than the AP domain has with the target, for example the P domain can have about 30% complementarity with the target while the AP domain can have substantially more complementarity, e.g. 50% to 100% complementarity.

Moreover, the degree of complementarity that provides detectable binding between the subject circular oligonucleotides and their respective targets, is dependent upon the conditions under which that binding occurs. It is well known that binding, i.e. hybridization, between nucleic acid strands depends on factors besides the degree of mismatch between two sequences. Such factors include the GC content of the region, temperature, ionic strength, the presence of formamide and types of counter ions present. The effect that these conditions have upon binding is known to one skilled in the art. Furthermore, conditions are frequently determined by the circumstances of use. For example, when a circular oligonucleotide is made for use in vivo, no formamide will be present and the ionic strength, types of counter ions, and temperature correspond to physiological conditions. Binding conditions can be manipulated in vitro to optimize the utility of the present oligonucleotides. A thorough treatment of the qualitative and quantitative considerations involved in establishing binding conditions that allow one skilled in the art to design appropriate oligonucleotides for use under the desired conditions is provided by Beltz et al., 1983, Methods Enzymol. 100:266-285 and by Sambrook et al.

Thus for the present invention, one of ordinary skill in the art can readily design a nucleotide sequence for the P and AP domains of the subject circular oligonucleotides which exhibits sufficient complementarity to detectably bind to its target sequence. As used herein "binding" or "stable binding" means that a sufficient amount of the oligonucleotide is bound or hybridized to its target to permit detection of that binding. Binding can be detected by either physical or functional properties of the target:circular oligonucleotide complex.

Binding between a target and an oligonucleotide can be detected by any procedure known to one skilled in the art, including both functional or physical binding assays. Binding may be detected functionally by determining whether binding has an observable effect upon a biosynthetic process such as DNA replication, RNA transcription, protein translation and the like.

Physical methods of detecting the binding of complementary strands of DNA or RNA are well known in the art, and include such methods as DNase I or chemical footprinting, gel shift and affinity cleavage assays and light absorption detection procedures. For example, a method which is widely used, because it is so simple and reliable, involves observing a change in light absorption of a solution containing an oligonucleotide and a target nucleic acid at 220 to 300 nm as the temperature is slowly increased. If the oligonucleotide has bound to its target, there is a sudden increase in absorption at a characteristic temperature as the oligonucleotide and target dissociate or melt.

The binding between an oligonucleotide and its target nucleic acid is frequently characterized by the temperature at which 50% of the oligonucleotide is melted from its target. This temperature is the melting temperature (T.sub.m). A higher T.sub.m means a stronger or more stable complex relative to a complex with a lower T.sub.m. The stability of a duplex increases with increasing G:C content since G:C base pairs have three hydrogen bonds whereas A:T base pairs have two. The circular oligonucleotides of the present invention provide additional hydrogen bonds and hence more stability since two binding domains are available for bonding to a single target nucleic acid, i.e. the P domain and the AP domain. Hence, the triplex formed by a circular oligonucleotide bound to its target nucleic acid should melt at a higher T.sub.m than the duplex formed by a linear oligonucleotide and a target.

Circular oligonucleotides bind to a nucleic acid target through hydrogen bonds formed between the nucleotides of the binding domains and the target. The AP domain can bind by forming Watson-Crick hydrogen bonds (FIG. 1). The P domain can bind to the target nucleotides by forming non-Watson-Crick hydrogen bonds (e.g., FIG. 1 and Table 1). When two nucleotides from different strands of DNA or RNA hydrogen bond by the base pairing rules defined herein, a base pair or duplex is formed. When a nucleotide from AP and a nucleotide from P both bind to the same target nucleotide, a base triad is formed.

Parallel domain base pairing with a complementary target strand of nucleic acid, is thermodynamically less favorable than Watson-Crick base pairing; however, when both parallel and anti-parallel pairing modes are present in a single molecule, highly stable complexes can form. Thus, two opposing domains of a circular oligomer form a complex with a central target, giving a triplex structure, or a triple helical complex, bounded by the two looped ends of the circle. For example, this arrangement can allow formation of up to four hydrogen bonds when two thymines bind to a target adenine and up to five hydrogen bonds when two cytosines bind to a target guanine.

Furthermore, because of the binding characteristics of the P and AP domains, the present circular oligonucleotides have a higher selectivity for a particular target than do corresponding linear oligonucleotides. At least two factors can contribute to this high selectivity. First, circular oligonucleotides of this invention bind twice to the same central target strand. Hence two domains are involved in selecting a target. Second, protonation of cytosine in a C+G-C triad is favored only when this triad forms and the additional proton gives the triad a positive charge. This positive charge can lessen the negative charge repulsions arising from the juxtapositioning of three phosphodiester backbones.

Protonation of C+G-C triads occurs most readily at low pH and formation of C+G-C triads is favored over formation of many other triads at low pH. Therefore, P and AP domains which are cytosine-rich more stably bind a complementary guanine-rich target at low pH than cytosine-poor P and AP domains bind a guanine-poor target. The skilled artisan can take advantage of the effect of protonation upon C+G-C triad formation to design circular oligonucleotides in accordance with the present invention whose selectivity for a target is enhanced if the pH of the hybridization reaction is known or can be adjusted. This is done simply by selecting a guanine-rich target and constructing cytosine-rich P and AP binding domains if the hybridization pH is low, or by selecting a guanine-poor target and constructing cytosine-poor P and AP binding domains if the hybridization pH is high. For these purposes a low pH is about 5.0 to about 6.8, and preferably about 5.5, whereas a high pH is about 7.0 to about 9.0, and for use in vivo preferably about 7.4. As used herein a cytosine-rich P or AP binding domain has about 2 to about 20 cytosines, and a guanine-rich target has about 2 to about 20 guanines. Conversely, a cytosine-poor P or AP binding domain has no more than one cytosine, while a guanine-poor target has no more than one guanine.

The circular oligonucleotides of the present invention can be constructed to include more than one P or AP binding domain to permit binding of the oligonucleotide to more than one target. The skilled artisan can also select target sites for such multiple-binding domain oligonucleotides which permit construction of cytosine-rich and cytosine-poor pairs of P and AP binding domains. By including a cytosine-rich pair of binding domains with a cytosine-poor